Sender: vheyndri AT rug DOT ac DOT be
Message-Id: <34E41E5B.3A7B@rug.ac.be>
Date: Fri, 13 Feb 1998 11:20:11 +0100
From: Vik Heyndrickx <Vik DOT Heyndrickx AT rug DOT ac DOT be>
Mime-Version: 1.0
To: DJGPP workers <djgpp-workers AT delorie DOT com>
Subject: a few proposed changes to alpha-980101
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Precedence: bulk

Hi workers,

I've made a few changes to the alpha-980101 distribution. I sent the
patch file to DJ, but you can have a look at it at URL
http://studwww.rug.ac.be/~vheyndri/djgpp-patches/orig_03.dif .
For every modification there is a good reason, so I sincerely hope that
every one of them is applied. If the reason for some change is not clear
by itself, please ask.

The differences where taken all at once by comparing two directory
trees.
Please use "patch -p1".

What:
* include/inlines/pc.h
  The promessed patches for the inport* macro's.

* for a few fsext related function it was necessary to turn 'int' into
ssize_t (which is also 'int' until further notice) in order to get a
consistent type passing.

* include/math.h
  Contained double prototypes for four functions.

* include/sys/segments.h
  (int)(short)0x8400 is 0xffff8400 which might eventually lead to a
problem because that is not valid selector value.

* a small but important change to MK_FOFF (the same reason as the
previous, but even more important).
I suggest also to move this macro to a header file (the same that
defines the dosmemput/*get functions), because it offers a correct and
consistent way to calculate a linear address from a (seg, ofs) pair.

* missing header file inclusions (only reported when compiled with
'-fno-builtin')

* the correct value for _dos_ds.

* _read and _write had wrong function headers in the *.c files (refer to
the *.h file)

* removed the masking for conventional memory linear addresses with
0xfffff. Those addresses are supposed to be correct in the first place
as a result of any calculation and if they are not, 0xfffff will only
make them valid conventional memory addresses, but bogus pointers. It
also removes the limititation which excluded the HMA (important to
TSR's). This way it will be easier to detect bogus code, since this will
now generate a proper #GP.

That it for now... :)

-- 
 \ Vik /-_-_-_-_-_-_/   
  \___/ Heyndrickx /          
   \ /-_-_-_-_-_-_/